Use 3 fields to create one dimension

[NathanS2022]

Hi, 

I have a doubt about the use of more than one field in a dimension. Previusly I use the load of 5 qvds in a concatenate table. In that situation, the field DIVISAO_VENDA was used in a dimension and works really fine. 

But, now, I had to use the load of qvds without  a concatenate table. So, the value of the dimension isn´t working how expected.

There is a way of concatenate only the field DIVISAO_VENDA between the qvds and use it as a dimension?

Below are images of how loading was done before and how it is being done now.

[marcus_sommer]

Why trying a different approach and not concatenating those files also by harmonizing the field-names? If the source-information is needed they could be added within an extra field - manually or with filebasename() or this information is already added by creating the qvd's ...

- Marcus

[NathanS2022]

Marcus, by harmonizing the field-names you mean give all the fields the same name?

If I do that, Qlik creates a synthetic key, which I´m trying to avoid.

I also try to create a unique table to deal with the field DIVISAO_VENDA. But doesn´t work. 

If there´s another approach, give me an example, please. 

Thanks at all

[marcus_sommer]

The fields within the single concatenated fact-table should be harmonized - means there is only one field for a date and one for an amount and one for a category and so on and not date_1 and date_2 and date_3. If a differentiating between the multiple sources is needed an extra source-field could be created which holds this information and which could be later used as dimension and/or a selection and/or a set analysis condition.

Important is to create a star-scheme as data-model - one fact-table with n dimension-tables. It's very easy to develop and it worked very well with a good performance. Everything else becomes mostly more complex and has usually no benefits else the opposite with more or less disadvantages against to the star-scheme. 

- Marcus

[NathanS2022 · Accepted solution]

Marcus, thanks for the explanation.

I solve it by concatenating the tables but keeping most of the fields with different names. So, now there´s a link beetwen tables by the DATA field and I still can use th others fields separaly.

My dimension is working because of the "junction" of the DIVISAO_VENDA field in the concatenate schema.